How to check neighbouring elements in a 2D array

I have a 2D array (~1000x1000) of intensity values that I would like to transform into an image to use as a mask for data processing in another program. I would like to mask out the dark lines and the corners:

What I am thinking is that for each element in the array, the neighbouring values (e.g. in a set radius) are checked for the proportion of zero values. If the proportion of zero values is above a set threshold, it should return a zero and otherwise it should return a one.

Happy to hear suggestions about other ways of going about this, and thank you in advance.

[KSSV]

To checl neighbouring elements read about knnsearch. What exactly you are epxecting? You want to make black lines inside the whitre region to white?

[Oliver Lindsay]

Basically I would like the black regions (lines and corners) to be solid black, and the white regions to be solid white.

Here is an example of a type of output I'd be looking for, noting that it doesn't have the corners cut out.

Answer 1

[Matt J]

load Image

BW=bwareaopen(~BW,100);

se=ones(7);

BW=medfilt2(BW,size(se));
BW=bwareaopen(BW,100);
BW=~bwareaopen(~BW,100);
BW=imclose(BW,se);

BW=conv2(BW,ones(5),'same')>0.5;

imshow(BW,[]); shg

[Oliver Lindsay]

Thanks!